Coal working face gas emission prediction based on optimization input factors and RVM

To improve the coal working face gas emission forecasting, an approach to determine the most suitable input factors of the coal working face gas emission forecaster is suggested based on the variance ratio testing method The necessary eliminations or adding of input factors are determined by F test for the forecasting variances before and after elimination or adding. After investigating on all cases, the most suitable input factor combination can be identified The measured samples of gas emission and some related factors at a coal working face are used as an example The input factors of a RVM(Relevance Vector Machine) based nonlinear forecaster are selected by the proposed approach It is shown that the forecasting results by the forecaster with the most suitable input factors are remarkably improved indicating that the proposed approach is feasible and effective, and the ability of nonlinear function approximation and generalization for prediction model of gas emission are further improved by the forecaster based RVM, and the mean prediction error are reduced to 1.74%.